18.228 gender and the Turing Test

From: Humanist Discussion Group (by way of Willard McCarty willard.mccarty_at_kcl.ac.uk>
Date: Mon, 20 Sep 2004 07:13:53 +0100

               Humanist Discussion Group, Vol. 18, No. 228.
       Centre for Computing in the Humanities, King's College London
                     Submit to: humanist_at_princeton.edu

         Date: Mon, 20 Sep 2004 07:09:55 +0100
         From: lachance_at_origin.chass.utoronto.ca (Francois Lachance)
         Subject: The Turns of Turing


In Humanist 18.201 Paul Oppenheimer suggest subscribers might wish to read
Sylvia Kelso's 1994 piece entitle The Silver Metal Imagination:
Blueprints for Changing Technology in Women's SF. The piece contains a
paragraph asking a set of questions. The paragraph in question is inspired
by Turing's thought experiment on the imitation game.


     Basically, an AI is a computer good enough to pass the test proposed
     by Alan Turing: that from another room you can't tell you're not
     communicating with a human. As with robots, SF has good and bad
     visions of such super-computers. Probably the classic bad case is HAL
     in Kubrick's 2001 (1968). A less famous good case is Mycroft Holmes,
     who helps run a revolution in Robert Heinlein's The Moon is a Harsh
     Mistress (1966). When AI's pass the Turing test, however, they raise
     all sorts of problems with, What is human? Is an AI just "a person in
     a box?" (bes shahar, 1991, 251). Can you be a person if you have
     microchips instead of cerebral lobes, and superconductors instead of
     synapses, and camera pickups for eyes? And what happens when you fall
     in love?


It is perhaps a tad odd that the silver metal imagination misses a very
important aspect of the imitation game as proposed by Turing. Turing's
imitation game is about gender ascription. Passing the test is very much
about passing. To be fair, what has popularly become know as the Turing
test is not necessarily what is reflected in the Turing text ("Computing
Machinery and Intelligence" published in 1950).

It is worth signalling this divergence between the commonplace of the
Turing test and the the thought experiment as described by Turing to the
influential subscribers of Humanist.The gender dynamic is cropping up in
discussions of Turing. Recently, the question of gender was very much part
of a discussion of Turing among other in a thread on the jill/txt blog
(See the entry date27/8/2004 "why are sex and computers conflated?").
Turing's proposition is not about the certitude of being (i.e. that
machines _are_ persons). It is about the accuracy of judging. He stated
the belief "that in about fifty years' time it will be possible to program
computers, with a storage capacity of about 10E9, to make them play the
imitation game so well that an average interrogator will not have more
than 70 per cent chance of making the right identification after five
minutes of questioning." Turing's belief is as much about machines as
about the relations between the sexes.

As one observer in the jill/txt discussions noted

The game is set up for the interrogator to fail when a machine is
involved. Whatever gender-sex the interrogator ascibes to the
machine-player, the interrogator is wrong.


The machine is neither a man or a woman. The interrogator looses either
way, unless the game is modified for the interrogator to be able to
provide a response of neither-nor.

The observer further notes that the of the exchange of roles (trying to
trip up the interrogator, trying to assist the interrogator and being the
interrogator) implies the possibility that the computer, the thinking
machine, can take the place of the interrogator. It is to be noted that
the interrogator is also in this version of the imitation game the
adjudicator, both the question asker and the decision maker. Neither
interrogation nor adjudication by a machine are explicitly explored in
Turing's essay. A similar undertheorizing of the phenomenology of
decision-making conditions the imagination of possibilities in the
concluding section of Turing's essay that of the learning machine. What
can a machine teach? How can a machine teach? How does a machine make
decisions? How do humans make decisions? How does an author know which
questions to substitute when? Turing himself in the turns of his essay,
especially his rereading of himself in the substitution of questions,
teaches us that teaching is about reading. A machine can teach us to read.
The machine produces readings. Humans have the ability to mimic reading
machines and the ability to learn gender.

-- Francois Lachance, Scholar-at-large http://www.chass.utoronto.ca/~lachance

A calendar is like a map. And just as maps have insets, calendars in the
21st century might have 'moments' expressed in flat local time fanning out
into "great circles" expressed in earth revolution time.
Received on Mon Sep 20 2004 - 02:24:22 EDT

This archive was generated by hypermail 2.2.0 : Mon Sep 20 2004 - 02:24:24 EDT